False Memories: Phenomena, Theories, and Implications Melanie Caroline Steffens Friedrich-Schiller-Universität Jena Silvia Mecklenbräuker ...

Page created by Gabriel Soto
 
CONTINUE READING
False Memories:

                          Phenomena, Theories, and Implications

                                   Melanie Caroline Steffens

                             Friedrich-Schiller-Universität Jena

                                    Silvia Mecklenbräuker

                                       Universität Trier

Address of corresponding author:

Melanie Steffens
Institut für Psychologie
Friedrich-Schiller-Universität Jena
Am Steiger 3, Haus 1
D-07743 Jena, Germany
E-mail: Melanie.steffens@uni-jena.de
Phone:+49-3641-945 111
Fax:+49-3641-945 112

Authors’ Notes
Please address correspondence to Melanie Steffens, Institut für Psychologie, Friedrich-
Schiller-Universität, Am Steiger 3, Haus 1, D-07743 Jena, Germany, e-mail:
Melanie.steffens@uni-jena.de, or to Silvia Mecklenbräuker, FB I-Psychologie, Universität
Trier, Universitätsring 15, D-54286 Trier, Germany, e-mail: mecklen@uni-trier.de. The
writing of this article was supported by a grand from the Deutsche Forschungsgemeinschaft
(DFG, Ste 938/6-1).

Word count: 7734
Steffens & Mecklenbräuker                                                 False Memories    2

                                              Abstract

       In recent years, there has been an explosion of research on false memories: the
subjective experience of remembering something if that something did apparently not happen
in reality. We review a range of findings concerning this phenomenon: False memories of
details and of whole events by adults and children, as well as false memories of words in
laboratory experiments (in the DRM paradigm). We also briefly discuss the converse
phenomenon: Evidence of forgetting or repression of significant events, and evidence of
recovered memories. Knowledge of both phenomena is needed for judging whether “new”
memories are false, recovered, or whether both options are possible. More general as well as
specific theories explaining false memories are discussed, and we close with implications for
practice.

Key words: False memory, DRM paradigm, source monitoring, fuzzy trace theory, repression
Steffens & Mecklenbräuker                                                      False Memories      3

                   False Memories: Phenomena, Theories, and Implications
       As an introduction to the practical relevance of the topic, think of a woman who has
been in therapy for many years because she has difficulties trusting men; she has panic
attacks; she has a very distant relationship with her father; and she suffers from disturbing
nightmares. At some point during that therapy, she “recovers” faint memories of early
childhood abuse by her father. She consequently breaks off all contact with her family of
origin. Can we be sure that she truly recovered old memories, as opposed to reconstructing
memories that are in line with her current beliefs and feelings? The self-help literature for
victims of abuse is very clear. For instance, “You must believe that your client was sexually
abused, even if she herself sometimes doubts it. … No one invents abuse. Neither children
nor women make up having been abused…” (Bass & Davis, 1992, p. 324, translated from
German). But what do experts on memory say? Can faint memories of early childhood be
false memories due to suggestions? How can such memories be induced? And how can
theories of memory explain this? Finally, what are the practical implications for situations
such as that in the above example? We start our overview of the false-memory literature (for
others, see Koriat, Goldsmith, & Pansky, 2000; Schacter, 1999) with a description of the
range of phenomena that are discussed under this topic because false memories are primarily
phenomena: astonishing findings.

                                              Phenomena
       A main reason why there was much less research on memory illusions than on
perceptual illusions until the middle of the 1990s is probably that it is tricky to obtain close
control over the stimulus that is subject to a memory illusion (Roediger, 1996). Illusions of
perception can be created and investigated in the laboratory within a minute. Memory
illusions probably arise mostly after longer latencies, and it is difficult to obtain objective
evidence as to what “really” happened. In these cases, discrepancies attest to the existence of
memory illusions. First, there are within-person discrepancies, such as reported for depression
patients who recall as straining and negative even events previously judged as pleasant.
Second, there are discrepancies between persons, as everyone has probably witnessed when
old couples get into verbal arguments about long-ago events (“The car was robbed when we
were on the way to the Black Forest.” – “No, we were not! That was the year before. The car
was robbed when we were on the way to Munich to visit your sister.” – “Darling, I am
positive that you are wrong because …”).
Steffens & Mecklenbräuker                                                  False Memories      4

                                     Evidence of False Memories
       If we assume that no one is lying on purpose in such dialogs, some distorted memories
must be involved here. It would, however, be very difficult to find out what actually happened
back then. Exceptions are memories for past things that are well documented. In a very nice
study using that approach, it was shown that people generally remember their good high-
school grades, but forget the bad ones and then overestimate them (Bahrick, Hall, & Berger,
1996). The researchers asked participants how well they did in school. After the interview,
they checked the grades in their high-school diplomas.
       What other approaches have been used to find out whether adults can have false
memories? An older thread of research along these lines, is research on eyewitness
testimonies. A strongly established memory phenomenon is that eyewitnesses often confuse
misleading post-event information with what they have witnessed. In the classic studies by
Loftus and her colleagues (for a review, see Loftus, Feldman, & Dashiell, 1995),
eyewitnesses who were asked whether they saw “the car run over the stop signal” were, in a
later questionnaire, more likely to indicate that they saw a stop signal when there had been
none, a phenomenon called misinformation effect (cf. McCloskey & Zaragoza, 1985, for a
discussion of the processes involved). For illustrative purposes, we present the study by
Zaragoza and Mitchell (1996). After having watched a movie of a burglary, misleading post-
event information was introduced to participants in a questionnaire with questions such as “At
the beginning of the scene, a young man dressed in jeans, a t-shirt and gloves entered the
house. Did he enter through the door?” (p. 295). Actually, the thief had not worn gloves. Each
piece of misleading information was presented for zero (control condition), one, or three
times. In the test phase, participants were warned that the questionnaire had contained
misleading information. They then did a source-monitoring test where they were asked
whether each piece of information (“The robber wore gloves”) was from the video or from the
questionnaire. The false alarm rate of details that had not been presented in the movie was
10% in the control condition, increased to 37% for false information presented once, and to a
remarkable 56% for misleading information presented three times (for theoretical accounts,
see below). Apparently, wrong details can quite easily be suggested to people and then be
incorporated into their memories of events (for a recent review, see Loftus, 2005a).
       There are many differences between suggesting details about a movie to someone and
suggesting whole events that allegedly happened to them. This latter phenomenon has also
been heavily researched in the last decade. A famous study conducted to test how susceptible
children are to false memories is the Sam-Stone experiment (Leichtman & Ceci, 1995). The
Steffens & Mecklenbräuker                                                     False Memories     5

authors’ aims were testing the roles of age, stereotypes, and suggestions for the formation of
false memories. Suggestions were comparable to the misleading postevent information
discussed above (also see White, Leichtman, & Ceci, 1997, for the effects of suggestions on
children’s reports). Stereotypes are of much interest in this context because child witnesses
might often (subtly) learn that a certain person is regarded as a “bad guy”. If this knowledge
influences their willingness to attest to bad things he did, this is of course a problem for
evaluating what child witnesses say. The third factor in the study, age, was considered
because, as one might expect, the reliability of child testimonies increases with their age (for
reviews, see Ceci & Bruck, 1993; Pipe, Lamb, Orbach, & Esplin, 2004). Children were tested
in day-care centres. The design of the study was a 2 (stereotype induction vs. no stereotype) X
2 (suggestion vs. no suggestion). In the weeks preceding the visit, children in the stereotype
induction condition heard stories about Sam Stone as a clumsy and bumbling person. Sam
Stone’s visit itself lasted only a few minutes. He came in, greeted, walked around, and left.
Afterwards, a torn book and a dirty teddy bear were discovered. Each child in the suggestion
conditions was interviewed repeatedly in the following weeks. Two questions were phrased in
a suggestive way (“Was Sam Stone happy or sad that he got that bear dirty?”). In the final
interview by a new interviewer, each child was asked whether Sam Stone soiled the bear and
ripped the book. Figure 1 shows the percentage of children affirming at least one of these
questions. Both the stereotype and the suggestions increased the number of false memories,
and false memories were more prevalent among four-year-olds than among six-year-olds.
False-memory rates of up to three fourths make something very clear about the reliability of
the memories of child witnesses: A four-year-old who has learned that a certain person did
something bad and who has repeatedly been interviewed in suggestive ways cannot be relied
on as a witness anymore. This is an explanation for the strange stories that child witnesses
sometimes relate: They want to be consistent with expectations, but if they don’t know much
about these expectations, their imagination can lead them astray. Younger children may show
a tendency to claim that all kinds of events have happened to them. For instance, if children
had drawn pictures of impossible events (flying to the moon on a rocket, swimming with
mermaids, etc.), they were more likely one week later to claim that such events had actually
happened to them in their past (Strange, Garry, & Sutherland, 2003). In short, only experts
should interview children who might be important witnesses, and they should do so following
specific rules (cf. Fisher & McCauley, 1995).

                                       Figure 1 about here
Steffens & Mecklenbräuker                                                     False Memories      6

       The Sam-Stone experiment is illustrative also because the researchers found a nice
way to avoid the frequent problem (alluded above) of having to investigate what happened in
the past: They included the “study phase” in the experiment (for a similar recent study with
positive and negative misinformation, see Memon, Holliday, & Hill, 2006). It would, of
course, be nearly impossible to expose children to a disturbing event and ask them 20 years
later what they remember of if. Other ways of controlling for the initial event are mentioned
further below. But first, we concentrate on studies that have suggested specific events to their
adult participants, only to test subsequently whether participants believed that these events
had actually happened to them (for a review, see Loftus, 2003). Interestingly, studies on false
memory “implantation” in adults were in part motivated by a famous story by Piaget (1962).
Piaget had a vivid memory of an attempted kidnapping that had allegedly happened early in
his childhood. “I can still see, most clearly, the following scene, in which I believed until I
was about fifteen. I was sitting in my pram, which my nurse was pushing in the Champs
Elysées, when a man tried to kidnap me. I was held in by the strap fastened round me while
my nurse bravely tried to stand between me and the thief. She received various scratches, and
I can still see vaguely those on her face…” (cited after Loftus, 2005b, p. 39). When he was
about fifteen, his former nurse confessed that she had made up the whole story. Piaget’s
example illustrated vividly the possibility that a repeated family story might lead a young
child to have a persisting false memory.
       Is it also possible to plant memories for nonexistent childhood experiences in adults?
A number of paradigms for false memory implantation have been developed (see Koriat et al.,
2000; Loftus, 2003). One example is the study by Hyman and colleagues (Hyman, Husband,
& Billings, 1995) including the event of spilling punch on the bride’s parents (for a more
recent example, see Ost, Foster, Costall, & Bull, 2005). In preparation, the researchers
collected several childhood events from students’ families. Relatives were asked not to talk to
the students about these details. Then, each student was interviewed and asked to describe as
many details as possible of 3-5 events that had actually happened in their childhood (a
camping holiday etc.). These events were intermixed with one standardized unlikely event
that the researchers had invented: playing wildly on a wedding, bumping into a bowl of punch
and spilling it over the bride’s parents. Participants were then dismissed with the hint that
they should try to, and would probably remember more details of the events in the next few
days, and were rescheduled for a second and third interview. At the end of the study, nearly
every fourth participant claimed to remember the fictitious event; they produced additional
Steffens & Mecklenbräuker                                                     False Memories    7

details of it; and when asked to pick from their list of childhood events the one that had
actually not happened, they could not confidently discard the punch story. Some were very
surprised during debriefing that this was the false event. Participant questioning revealed
interesting differences between those participants who bought into the false event and those
who did not. Those who held general memories that were consistent with the false event were
more likely to believe that it had happened and to produce false memories (in the example:
wild children who had been on weddings). In addition, those who had specific reasons to
dismiss the story did not produce false memories (e.g., “it must have been my brother”).
       “Planting” false memories in around 25% of the participants is in line with other
experiments on false childhood memories. This proportion is still increased if more
information is presented that adds to the credibility of the event: When faced with a (merged)
picture of them in a hot-air-balloon, 60% of participants produced false memories of the
balloon ride (Lindsay, Hagen, Read, Wade, & Garry, 2004; Wade, Garry, Read, & Lindsay,
2002; for a review on recent work with this doctored photo procedure, see Garry & Gerrie,
2005). Even when the photo was only associated with (but not depicting) the suggested
pseudoevent (school classes’ group photos from the years of the to-be-recalled school-related
childhood events), two-thirds of the participants developed false memories. As discussed by
Hyman et al. (1995), it is a typical feature of childhood memories that they are part of the
family history, revolving around pictures with accompanying stories. Therefore, it must be
very difficult to disentangle true memories of the events themselves from memories of the
pictures and the respective often-told stories, all the more so because childhood memories are
often blurred as opposed to rich and detailed.
       These findings prove that false memories of childhood events do exist. Can this
evidence still be refuted as irrelevant to recovered memories of childhood abuse? People who
have tried to do so (mostly therapists, who often take a different stance in this discussion than
memory researchers do, see Ost, 2003; Pope, 1996) have argued that false memories of not
too unpleasant, single events are very different from memories of repeated abuse by relatives.
This is certainly true (whereas, as a side note, it is not so clear why different processes should
be at work for false memories differing in valence etc.). The argument of interevent
differences has been taken up by researchers who have, in turn, designed experiments to test
whether false memories of unpleasant events can be created. They searched for an event that
is very unpleasant; where a child would feel helpless; and where those parts of her or his body
are involved that are usually not touched by others. Pezdek and Hodge (1999) investigated
whether children create false memories of receiving a rectal enema. Being lost in a shopping
Steffens & Mecklenbräuker                                                     False Memories      8

mall was used as a control event in the same study. Whereas 14 of the children tested created
false memories of being lost, but not of the hospital visit, this pattern was reversed for only
one child. This shows that differences between events influence the probability of false-
memory creation: When the plausibility of an event is subjectively low, apparently false
memories do not emerge (also see below).
       Confidence estimates regarding unlikely events can be increased by asking students to
vividly imagine the respective situations, an effect first reported by Rita Anderson (1984) and
later termed imagination inflation (Garry, Manning, Loftus, & Sherman, 1996). In the study
phase of the Garry et al. study, participants ticked each event on a list as to the probability
with which it had (not) happened to them during childhood (such as finding a dollar bill in a
parking lot, dialling an emergency number). Two weeks later, they were asked to imagine in
detail some of the events. Then, the experimenter pretended that she had made a mistake in
coding the previous lists and asked participants to please fill them out again. As it turned out,
those events that participants had imagined were now judged to have really happened with a
higher probability than before. However, a distinction should be drawn between confidence
estimates about fictitious events, beliefs about such events, and false memories (see Smeets,
Merckelbach, Horselenberg, & Jelicic, 2005). Another criticism of imagination inflation
studies is that there is, again, no control over which events happened in participants’
childhood and which did not. There is, after all, the possibility that some participants forgot
all about finding a dollar bill, so only when they pictured the event in some detail were they
reminded of the original event. Several studies indicate, however, that with multiple acts of
imagination, people can be led to believe that they performed an unlikely action (e.g., kissing
a magnifying glass; sitting on a dice) (e.g., Goff & Roediger, 1998; Thomas, Bulevich, &
Loftus, 2003).
       An alternative way to gain control over the past event is investigating false memories
of impossible events. Spanos and colleagues (see Spanos, 1996) successfully planted a false
memory that allegedly occurred one day after birth (a coloured mobile hanging over the crib).
Researchers have also investigated false memories of past lives that were created under
hypnosis (Spanos, Menary, Gabora, DuBreuil, & Dewhirst, 1991). These memories were
consistent with suggestions (such as the proportion of children having been abused back then)
and with American students’ knowledge rather than historic evidence (such as: recalling
having lived in Germany in the 16th century). Other researchers have tried to explain false
memories of alien abductions (see Banaji & Kihlstrom, 1996; McNally et al., 2004). Such
findings indicate that event plausibility is not a necessary condition for coming to belief that
Steffens & Mecklenbräuker                                                    False Memories     9

an event happened when methods are used that increase schematic knowledge and/or
document the false events with authority (e.g., reading accounts of “witnessing a demonic
possession”: Mazzoni, Loftus, & Kirsch, 2001). Using a variation of the doctored photo
procedure (see above), Strange, Sutherland, and Garry (2006) found that children (6- and 10-
year-olds) showed comparable rates of false memories for a higher plausibility event (taking a
hot air balloon ride) and a lower plausibility event (having a cup of tea with Prince Charles).
In short, false memories of whole and even implausible events can be planted not only in
children, but also in adults.

                                       Table 1 about here

                   Creating False Memories in the Laboratory: The DRM Paradigm
       How can false memories be efficiently investigated if there are many differences
between the events that are subject to false memories, and if every study used only one or a
handful of different events? For demonstration purposes, have a quick read through the list of
words presented in Table 1.
       False memories were taken into the laboratory by Deese (1959) who obviously had a
marketing problem: No one ever cared about his research until Roediger and McDermott
(1995) re-invented the paradigm. They obviously did a much better job selling it and,
consequently, it is now called the DRM paradigm (following a suggestion made by Endel
Tulving). For demonstration purposes, mark those words in Table 2 below that you have
previously read in Table 1 (without going back to Table 1, of course).
       In the DRM paradigm, lists of words are presented that include the first 12-15
associates for a given nonpresented critical word (critical lure, or prototype). Participants are
instructed to memorize the presented words for a later test, usually recall and/or recognition.
Figure 2 shows typical results (data from our laboratory) in a recognition test. It was
counterbalanced which list had been presented to a given participant. During recognition, list
words and critical words from all lists were presented. The false alarm rate for the nonstudied
critical words was much higher than that for other distractors. You may check for yourself
that none of the words in Table 2 was contained in Table 1. Generally, false recall and
recognition rates are fairly high for the nonstudied critical words (see Roediger & Gallo,
2003, for an overview). Moreover, the critical words were recognized with a high confidence,
and when remember/know judgments (Tulving, 1985) were required, true and false
recognitions were equally likely to give rise to remember responses (e.g., Roediger &
Steffens & Mecklenbräuker                                                  False Memories    10

McDermott, 1995). There is a tremendous amount of research on false memories in the DRM
paradigm (see e.g., Gallo, 2006; Roediger & Gallo, 2003, for reviews) and we will only
mention some interesting findings.

                                      Figure 2 about here

       There is evidence that false memory can be based largely on automatic processing and
is amenable to only limited conscious control. For example, Dodd and MacLeod (2004)
showed that mere exposure to DRM lists was sufficient to create a false memory: They
presented DRM lists as coloured words in a Stroop test. Naming colours reduced accurate
memory for list words as compared to reading coloured words, but false memory remained
high for critical words. Furthermore, false memory is difficult to eliminate. For example, false
recall is not reduced by directed forgetting instructions (e.g., Kimball & R. A. Bjork, 2002;
Seamon, Luo, Shulman, Toner, & Caglar, 2002), and young adults can take advantage of
explicit warning instructions given prior to encoding to reduce but not to eliminate false
memories (e.g., Watson, McDermott, & Balota, 2004; Westerberg & Marsolek, 2006). And a
last intriguing finding: False memories are remarkably persistent. For example, Toglia,
Neuschatz and Goodwin (1999) found that false recall rates remained high over a three-week
period, whereas recall of studied words revealed the typical decrement. In short, the DRM
paradigm allows for the easy and reliable elicitation of false memories in the laboratory.

                                      Table 2 about here

                                   Evidence of Repressed Memories
       Even though this overview is primarily focused on false memories, the other side of
the coin needs to be presented in order to arrive at a complete picture when we come to the
implications (also see Gleaves, Smith, Butler, & Spiegel, 2004). Memory researchers have
often claimed that the very problem with traumatic experiences is not that they are repressed
and forgotten, but that they remain vivid and cannot be forgotten (Loftus, 2003). Indeed, the
disturbing vividness with which pictures of the traumatic event are present is part of the
definition of the Post-Traumatic Stress Disorder, PTSD (see DSM IV). Therapists who
believe in repression often follow a Freudian tradition. Roughly, according to this theoretical
stance, memories that cannot be assimilated into existing schemas (such as those of a
normally loving father doing bad and wrong) are fearfully repressed into unconsciousness. In
Steffens & Mecklenbräuker                                                    False Memories     11

the course of therapy, when they appear less threatening, they can be recovered (cf. Loftus,
1993).
         The first relevant question thus is whether there is evidence that very significant and
emotionally upsetting events can be forgotten. Most people hold a meta-memory belief that
emotional events are remembered better than emotionally neutral ones, even though, as
research shows, the relationship between emotion and memory is a very complicated one (cf.
Klauer, 2000, for an overview). In contrast to the belief that emotional content and personal
relevance hinder forgetting, there is clear evidence to the contrary. Waagenaar and
Groeneweg (1990) showed that people who had been in a concentration camp during World
War II had, in the 1980s, forgotten many of the horrible things that had been done to them
and that they had been able to report soon after the war (but see McNally, 2005).
         Even more relevant to the false-memory debate are findings by Meyer Williams
(1994) on abused children. She contacted as many of these people who had been, according to
hospital files, brought into the emergency ward when they were children at the beginning of
the 1970s with severe physical harm resulting from abuse. During extended in-depth
interviews, she probed them for memories of that abuse. As one would expect, the probability
of remembering depended on the interviewee’s age at the time of abuse. Strikingly, however,
there was a considerable proportion who indicated no memory of the incidence at all, 38%
overall; even 25% of those who had been between seven and twelve years old when the
incident was reported. The elegance of this study derives from the neat solution of the
difficulty to gain control over the prior event: This was obtained by consulting the hospital
files that documented the childhood abuse. A few problems with the study should, however,
be discussed (cf. also Loftus, Garry, & Feldman, 1994). For instance, there is no objective
evidence that participants really could not remember the abuse: They simply might have
refrained from telling a stranger about it. This is unlikely, though, because these participants
told the interviewer other things where at least as much reluctance might have been expected.
In addition, it could be that these children never understood the events as sexual abuse: So
instead of forgetting them, they never encoded them as such. Nevertheless, in a nutshell, we
can conclude that disturbing events are not shielded from forgetting.
         The second question relevant to evidence of repressed memories is whether there is a
difference between repression and “normal” forgetting, as Freud and others would claim. In
other words, is there something in memory that resembles an active inhibition mechanism,
reducing the activation level of a representation and thus lowering its accessibility? At least
three research threads are relevant here. The first is the large literature on directed forgetting
Steffens & Mecklenbräuker                                                    False Memories     12

(for reviews, see E. L. Bjork, R.A. Bjork, & M. C. Anderson, 1998; MacLeod, 1998). Briefly,
what is shown is that words designated “to be forgotten” after study are recalled with a lower
probability than control words. They are recognized quite well, however, suggesting that
specifically their accessibility is lowered (for an alternative inhibition-free account of directed
forgetting, see MacLeod, Dodd, Sheard, Wilson, & Bibi, 2003; Sheard & MacLeod, 2005).
The second thread of research shows that rehearsing something lowers the probability of
recalling related things. For instance, repeated rehearsal of half of the pictures in a given
sequence of events (Koutstaal, Schacter, Johnson, & Galluccio, 1999) lowers the probability
of recalling those pictures that were not rehearsed (while at the same time, of course,
increasing the probability of recalling those that were rehearsed). It thus seems as if
remembering something represses things related (see Bäuml, this issue). Finally, a third
research thread shows direct evidence of repression (M. C. Anderson & Green, 2001). In a
very original study, in a paired-associate learning paradigm, participants were asked to learn
arbitrary pairs of words (such as “radio-lion”). In a second phase of the experiment,
participants were given the cue (“radio”) and asked to respond with the target (“lion”).
However, for a designated subset of the word pairs (repressor pairs), they were asked to avoid
any thoughts of the target when given the cue. In the final test phase, the original targets were
to be reproduced in response to the cues. The more often the repressor cue had been presented
in phase two, the less likely participants were to recall the target. More strikingly, even given
semantic cues (“four-legged animal”), they produced the target with a lowered probability
when the repressor cue had been presented repeatedly in phase two. Thus, it appears that
those targets themselves had been inhibited, not only the cue-target relation. Taken together,
the findings reported in this section demonstrate that first, traumatic childhood events can be
forgotten and second, that there is also some evidence for a repression mechanism in memory.

                            Theoretical Accounts of False Memories
       Can memory theories explain false memories? Do they need to be modified in the
light of findings on false memories? Do we need to develop new accounts?

                                     General Principles of Memory
       On a general level, we believe that traditional views of memory can easily account for
new findings on false memories (see also Koriat et al., 2000; Schacter, 1999). A first principle
highlighted in false-memory research is the reconstructive nature of memory that was already
stressed by Bartlett (1932): Events are not stored in memory as are pictures on a videotape.
Instead, events are reconstructed in the light of present views, using available schemas, that is,
Steffens & Mecklenbräuker                                                   False Memories     13

knowledge structures (for a classic review on schema-based effects on memory, see Alba &
Hasher, 1983). In fact, such schemas guide not only the retrieval of events, but even their
storage. For instance, Dunning and Sherman (1997) showed that a sentence like “The rock
star was upset about the amount of alcohol served at the party” was falsely remembered as
“The rock star was upset that so little alcohol was served at the party”, whereas the same
sentence with a nun as the subject was falsely remembered as “The nun was upset that so
much alcohol was served at the party”. More generally, inferences that are necessary for
understanding events are made immediately, and after a delay, they cannot be disentangled
from information directly given (Kintsch, 1988). In other words, a second reason why false
memories of events may occur is that ambiguous situations are interpreted in terms of existing
schemas, and they are then stored accordingly. Even in the absence of further distortions,
these events would be recalled falsely (for a review on the influence of schemas on false
memories, see Hirt, Lynn, Payne, Krackow, & McCrea, 1999).
       A third aspect of traditional memory theories that is well in line with findings, even
though this time with findings on recovered memories, is the broad consensus on fluctuations
in recall. In Tulving’s (1983) words, this is the difference between availability (something is
stored in memory) and accessibility (it may not be accessible under the given circumstances).
With powerful retrieval cues (such as the huge Göteborg ferry in a German harbour),
someone who has always claimed to have never been in Sweden may recall a day-trip there
fifteen years ago. Sleeping in a tent for the first time in 20 years, a man may recall that he was
molested by a priest during a camping trip. Even laboratory research shows fluctuations in
recall over the course of minutes: Hypermnesia is the phenomenon that words from a list that
were not recalled during the first recall test are recalled during a second test (with no
intervening second study phase) (e.g., Burns, 1993). One major determinant of accessibility
seems to be the whole study context, the interactive context (that influences the meaning of
what is studied) as well as the incidental context (cf. Baddeley, 1982), that is, everything
present during study, including sights, odors, and even own emotions (Bower, 1992). In line
with the encoding specificity hypothesis (Tulving & Thomson, 1973) and the transfer-
appropriate processing approach (Bransford, Franks, Morris, & Stein, 1979), aspects of the
whole context are encoded within a memory trace, and the more of these aspects are present
(or even imagined), the higher is the probability that the memory trace is activated. These
theoretical considerations imply that things long forgotten can be remembered if someone
extendedly and repeatedly thinks about them or imagines them, as is often the case with
childhood events during therapy. In short, memory theories from an information-processing
Steffens & Mecklenbräuker                                                     False Memories      14

perspective, not only those from a Freudian stance, predict that the recovery of memories
should be possible. (Every one can try for themselves by reconstructing a childhood song or
poem that is, during the first recall, hard to come up with, but where further words and lines
will be recalled on subsequent trials.) Several more specific accounts of false memory have
been brought forward.

                                 The Fluency-Misattribution Perspective
       False memories may result from the misattribution of processing fluency. According
to this view (e.g., Jacoby, Kelley, & Dywan, 1989), the subjective experience of familiarity
results from the unconscious attribution of fluent processing to the past (for a refinement of
this approach, called discrepancy-attribution hypothesis, see Whittlesea, Masson, & Hughes,
2005; Whittlesea & Williams, 2000). When fluency due to the previous presentation of a
stimulus is attributed to the past, veridical recognition results. However, other factors can lead
to fluent processing. In that case, fluency is misattributed to the past (i.e., to an incorrect
source) and an illusion of familiarity results. For example, imagining childhood events may
increase participants’ familiarity with these events (but see Mazzoni & Memon, 2003).

                                   The Source-Monitoring Framework
       Findings on misleading postevent information, including those where specific
childhood events were suggested to participants and those on imagination inflation, can all be
explained within a source-monitoring framework (Johnson, Hashtroudi, & Lindsay, 1993;
Johnson & Raye, 2000). The general tenets of this framework are as follows. We need to
distinguish between the content of a memory trace (“what” we know) and its source (“where”
we know it from). The source may be forgotten independent of the trace. You may know that
someone in the discussion mentioned the pope, but you forgot who it was. A source is not a
fixed part of a memory trace, like a tag attached to it. Instead, source attributions are decision
processes. For faint memories, they can be very difficult. Source-monitoring decisions can be
made between external sources (“Did you see on TV that they opened the Berlin wall or did
you hear it on the radio?”), between internal sources (“Did you state this brilliant argument in
the discussion or did you only imagine doing it?”), or between an internal and an external
source (“Did you state this brilliant argument or was it your friend?). This distinction between
internal and external sources emphasizes the origin of the information (the self versus external
source as origin), and not the public versus private status of the information (see Johnson et
al., 1993, Footnote 1). An example of a lack of source monitoring is described in a novel by
Jana Hensel (2002, Zonenkinder), who, as a teenager, attended the street marches in Leipzig
Steffens & Mecklenbräuker                                                    False Memories     15

that preceded the German unification: “I cannot tell anymore what I saw with my own eyes
and what I saw, at the same evening and then time and time again, in the Tagesthemen [daily
news]” (p. 12). Another recent example is the finding that over half of an undergraduate
sample reported the existence of a memory whose origin, read or dreamt, was uncertain to
them (Kemp & Burt, 2006). Faint memories are, according to the source-monitoring
framework, attributed to external sources when they are rich in perceptual details (“If I
remember my friend’s voice and her happy face, it was probably her who mentioned this
brilliant argument”). In contrast, attributions to internal sources become more likely if one
remembers the cognitive processes involved (“I remember how difficult it was to come up
with the argument”). The main point to keep in mind is, however, that general knowledge
such as meta-memory beliefs guide the decision process as to what was imagined and what
happened. From this stance, therapists have a lot of responsibility because they are seen as
authorities with regard to mental and emotional processes. If such an authority strongly and
repeatedly suggests that an adult must have transgressed a woman’s boundaries when she was
a child, given her current emotional status, and if that therapist even asks the client to
repeatedly imagine in detail what might have happened (cf. Garry et al., 1996) we can derive
from the source-monitoring framework the prediction that it will become increasingly
difficult for the client to accurately tell childhood memories from the respective imaginations.
In sum, the source-monitoring framework can account for many findings on false memories:
For instance, something that is really a mental picture from an imagination may be mistaken
for a childhood memory. And when people claim to remember details about events that never
happened, they may in fact remembering features of other, actually perceived events (Lyle &
Johnson, 2006).
       The source-monitoring framework is closely related to the fluency-misattribution
perspective, but emphasizes misattributions of specific characteristics (i.e., sensory details)
instead of global familiarity. Respective results were obtained by Thomas et al. (2003),
indicating that imagination inflation with few sensory details can be better explained by a
misattribution of global familiarity, whereas imagination inflation with many sensory details
is the larger effect and is more consistent with the source-monitoring framework (see also
Mazzoni & Memon, 2003).

                                  The Activation-Monitoring Account
       The Activation-Monitoring Account (AMA) was developed by Roediger and his
colleagues (e.g., Roediger, Watson, McDermott, & Gallo, 2001) to explain false memory
Steffens & Mecklenbräuker                                                     False Memories    16

findings from the DRM paradigm. The operation of two opponent processes is posited to
account for dissociative effects of some study variables (e.g., study time or number of study
trials) on accurate and false memory: activation of the critical words, usually during study,
and (source) monitoring or memory editing, usually during testing. During the study phase,
item-specific aspects of each list word are processed that increase its distinctiveness, and the
relations between the words are processed and activate also the critical word (see Hunt &
Einstein, 1981, for item-specific vs. relational processing). The representations of critical
words are activated either consciously by elaborative processing or automatically through the
spread of activation within an associative network (e.g., McDermott & Watson, 2001). At
test, participants must distinguish between activation resulting from actual list presentation
and internal activation. They may mistake the internal activation of the critical word’s
representation for its actual occurrence at study, thereby making a source-monitoring error.
The role of activation processes is strengthened by the regression analysis finding that the
strength of backward associations from the list items to the critical word is the best predictor
of false memories (Roediger et al., 2001; see also Anaki, Faran, Ben-Shalom, & Henik, 2005,
for the relevance of backward association strength). In other words, on the one hand, the
stronger the list items activate the critical word, the higher is the probability of false recall and
recognition. On the other hand, the probability of false memories decreases if monitoring
becomes easier, for instance, when lists are well learned (e.g., McDermott & Watson, 2001).
And the probability of false memories increases if monitoring becomes more difficult, for
instance, when not enough source information was encoded because participants were
stressed before study (Payne, Nadel, Allen, Thomas, & Jacobs, 2002). False memories are
even spontaneously recovered (after lists were well learned and false recognition rates were
greatly reduced) when participants were rushed or delayed, thus making monitoring more
difficult (Seamon et al., 2006).
       The characterization of monitoring processes in AMA borrows notions from the more
general source-monitoring framework. It is important to note that other theoretical accounts of
monitoring processes exist (e.g., in Fuzzy Trace Theory) which suggest similar processes to
those outlined in the source-monitoring framework but also propose some differences (for
details, see e.g., Gallo, Bell, Beier, & Schacter, 2006; Odegard & Lampinen, 2006).

                                           Fuzzy Trace Theory
       Fuzzy Trace Theory (FTT) is a theory on an intermediate level of abstraction (e.g.,
Brainerd & Reyna, 2002, 2004). It incorporates more detailed assumptions about processes
Steffens & Mecklenbräuker                                                   False Memories     17

underlying false memories than the general memory theories discussed above, but is wider in
scope than the AMA. The main tenets of the theory are that in parallel, two memory traces are
established. Verbatim traces represent the surface details of physical stimuli, whereas gist
traces represent the meaning or theme of the stimuli. Accurate memory of studied items is
driven largely by verbatim traces, whereas false memory is based predominantly on gist
traces. Verbatim traces decline faster than gist traces. Thus, false reports can display sleeper
effects, waxing over time even as true reports are waning. FTT is similar to schema theories
but includes more detailed assumptions, and the latter postulate a serial relation between
verbatim and gist traces: Verbatim traces are either lost or become integrated with schematic-
gist information. Like the AMA, FTT is an opponent process theory: False memory is
increased by a generation process that produces gist traces at study and is decreased by a
memory editing process that monitors recollection at test. Two predictions that have often
been confirmed empirically are that the probability of false memories increases if the gist is
repeated (as happens in the DRM paradigm) and if memory declines (e.g., with time passing).
       Whereas the FTT and the AMA postulate different types of monitoring processes, they
also share many features (e.g., they are two-process theories, and relational encoding
enhances the activation of the critical word/gist processing). The main difference is that
according to the AMA the representation of the critical word contains both semantic and
surface features, whereas gist traces corresponding to critical words include only semantic
information. The empirical evidence is still mixed (e.g., Cabeza & Lennartson, 2005; Coane,
McBride, Raulerson, & Jordan, 2006; McBride, Coane, & Raulerson, 2006). The accounts
may be reconciled by assuming that under some conditions, false memories may be mainly
due to associative activation, under others, they may mainly rely on meaning-based
information (cf. Hutchison & Balota, 2005).
       The similarities among the theoretical accounts we outlined by far outweigh the
differences. They complement rather than exclude each other. As a consequence, many false
memory phenomena are compatible with several approaches. The different accounts
emphasize different processes that all contribute to false memories, or even postulate similar
processes using different terminologies. Additional differences concern details of the
postulated processes that may very well depend on the specific conditions under which a
false-memory phenomenon occurs.
Steffens & Mecklenbräuker                                                   False Memories     18

                                   Constructive Memory Framework
       An integrative framework that incorporates most of the just described processes (e.g.,
constructive and reconstructive processing, item-specific vs. relational processing, spreading
activation, source monitoring) and may be seen as a unifying account is Schacter, Norman,
and Koutstaal’s (1998) neuropsychologically informed Constructive Memory Framework
(CMF). Representations of new experiences are conceptualized as patterns of features, and
retrieval involves a process of pattern completion in which a subset of the features are
reactivated, and activation spreads to the rest of the constituent features. Accurate
representations of past experiences require that features (e.g., perceptual, semantic,
contextual) comprising a specific episode must be linked together at encoding to form a
bound or coherent representation. Source-memory failures can result from inadequate feature
binding or from insufficient source information in the bound representation. Memory errors
can also result from insufficient pattern separation at encoding. If items (or episodes) overlap
with one another (as happens in the DRM paradigm), individuals may show good memory for
what the items have in common (i.e., gist information), but impaired recall of item-specific
information. At test, memory distortions can arise when people construct a retrieval cue that
is not fully consistent with information in the target trace. For example, people may use
schematic knowledge and/or information that is present in the test environment to construct
cues. Memory distortions can arise when schematic knowledge or physical retrieval cues (cf.
encoding specificity) fail to accurately describe a particular episode. And when the pattern
completion process produces a match, a decision must be made whether the information
delivered to awareness constitutes an episodic memory, as opposed to a fantasy or thought.
Memory errors can result from setting a lax criterion in this source monitoring decision.
       One problem with CMF may be that most false memory phenomena cannot be
understood unambiguously. For example, the DRM illusion may be attributed to pattern-
separation failure at encoding or a source-monitoring error at test. Closer specification (along
with empirical tests) is needed of the conditions when a given false-memory phenomenon
may be better explained by one or the other mechanism posited in CMF, and by more general
or more specific principles.

                                             Implications
       We can sum up the above review in the following words: There is abundant evidence
of false memories, not only in children, but also in adults. This evidence comes from a range
of research paradigms. There are rather general as well as quite specific theories brought
Steffens & Mecklenbräuker                                                    False Memories     19

forward to explain these findings. A first implication for future research is that researchers in
the field of false memories should now be more concerned with testing novel hypotheses that
bring forward the theoretical debate, instead of focusing on empirical phenomena.
       A second implication of our review is: Experts cannot decide with certainty whether a
given specific memory is “recovered” or “false” (cf. Gleaves et al., 2004). As we have
discussed, there are some things that therapists or interviewers should definitely avoid (most
of all, suggestions and instructions to imagine). Neuropsychologists try to discover correlates
of true, but not false memories (e.g., Nessler & Mecklinger, 2003; Okado & Stark, 2005;
Schacter & Slotnick, 2004; Slotnick & Schacter, 2004;). Thus, true and false memories could
one day be reliably distinguished on the basis of neurophysiological activity.
       Even though a few years ago, researchers have warned that the DRM paradigm is of
little relevance to the false-memory debate (Freyd & Gleaves, 1996), by now there is some
evidence that there may indeed be a connection. Some studies compared the rates of false
memories elicited in the DRM paradigm in control participants and in various clinical groups.
Women with PTSD after childhood sexual abuse (Bremner, Shobe, & Kihlstrom, 2000) and
women reporting recovered memories of childhood sexual abuse (Clancy, Schacter, McNally,
& Pitman, 2000) showed increased rates of false memories. Most importantly, Geraerts,
Smeets, Jelicic, van Heerden, and Merckelbach (2005) were able to show that the finding
obtained by Clancy et al. (2000) even held when trauma-related DRM lists were used, that is,
emotionally distinctive material (e.g., “rape, police, beating, child, violence” for the critical
word “assault”). These findings suggest that this population might be especially vulnerable to
memory distortions, possibly due to a source-monitoring deficit. Indeed, women reporting
recovered memories are also more vulnerable to forgetting prior remembering (“forgot it all
along” phenomenon, Geraerts et al., 2006).
       A few years ago, the “memory wars” split psychology into therapists believing their
clients’ reports (we are quite certain that this is, on the one hand, what therapists should do)
and researchers enquiring skeptically the validity of such reports (we are positive that this is,
on the other hand, what one should ask of researchers). In the popular-press book “The myth
of repressed memory”, Loftus wrote: “Something has gone wrong with therapy, and because
that something has to do with memory, I find myself at the center of an increasingly bitter and
fractious controversy. On the one side are the ‘True Believers’ who insist that the mind is
capable of repressing memories… On the other side are the ‘Skeptics’…” (Loftus &
Ketcham, 1994, p. 31). Those whom she calls, not too flattering, the ‘True Believers’, were so
outraged about the decisiveness with which she questioned the existence of recovered
Steffens & Mecklenbräuker                                                 False Memories   20

memories that they sent her incredibly fierce letters: “Please consider your work to be on the
same level as those who deny the existence of the extermination camps during World War II”
(Loftus & Ketcham, 1994, p. 36). It seems that by now, some of the steam has cooled down,
and the not-very-original answer probably lies in the middle: Memory theories predict, and
findings prove, that memories, even if held confidently, may be false. Memory theories have
also predicted since the 1970s, and older as well as newer findings show, that things that once
appeared forgotten can possibly be recovered under other circumstances. As William James
put it, “Memory is then the feeling of belief in a peculiar complex object… the re-recollected
past and the imaginary past may be much the same… there is nothing unique in the object of
memory…” (James, 1890/1950, p. 652).
Steffens & Mecklenbräuker                                                 False Memories   21

                                            References
Alba, J. W., & Hasher, L. (1983). Is memory schematic? Psychological Bulletin, 93, 203-231.
Anaki, D., Faran, Y., Ben-Shalom, D., & Henik, A. (2005). The false memory and the mirror
        effects: The role of familiarity and backward association in creating false
        recollections. Journal of Memory and Language, 52, 87-102.
Anderson, M. C., & Green, C. (2001). Suppressing unwanted memories by executive control.
        Nature, 410, 366-369.
Anderson, R. E. (1984). Did I do it or did I only imagine doing it? Journal of Experimental
        Psychology: General, 113, 594-613.
Baddeley, A. D. (1982). Domains of recollection. Psychological Review, 89, 708-729.
Bahrick, H. P., Hall, L. K., & Berger, S. A. (1996). Accuracy and distortion in memory for
        high school grades. Psychological Science, 7, 265-271.
Banaji, M. R., & Kihlstrom, J. F. (1996). The ordinary nature of alien abduction memories.
        Psychological Inquiry, 7, 132-135.
Bartlett, F. C. (1932). Remembering: A study in experimental and social psychology.
        Cambridge: Cambridge University Press.
Bass, E., & Davis, L. (1992). Trotz allem. Wege zur Selbstheilung für sexuell mißbrauchte
        Frauen (The courage to heal. A guide for women survivors of child sexual abuse).
        Berlin: Orlanda Frauenverlag.
Bäuml, K. H. (in press). Making memories unavailable: The inhibitory power of retrieval.
        Zeitschrift für Psychologie.
Bjork, E. L., Bjork, R. A., & Anderson, M. C. (1998). Varieties of goal-directed forgetting. In
        J. M. Golding & C. M. MacLeod (Eds.), Intentional forgetting: Interdisciplinary
        approaches (pp. 103-137). Mahwah, NJ, USA: Lawrence Erlbaum Associates, Inc.
Bower, G. H. (1992). How might emotions affect learning? In S.-A. Christianson (Ed.), The
        handbook of emotion and memory: Research and theory (pp. 3-31). Hillsdale, NJ,
        USA: Lawrence Erlbaum Associates, Inc.
Brainerd, C. J., & Reyna, V. F. (2002). Fuzzy-trace theory and false memory. Current
        Directions in Psychological Science, 11, 164-169.
Brainerd, C. J., & Reyna, V. F. (2004). Fuzzy-trace theory and memory development.
        Developmental Review, 24, 397-439.
Bransford, J. D., Franks, J. J., Morris, C. D., & Stein, B. S. (1979). Some general constraints
        on learning and memory research. In L. S. Cermac & F. I. M. Craik (Eds.), Levels of
        processing in human memory (pp. 331-354). Hillsdale, NJ, USA: Lawrence Erlbaum
        Associates, Inc.
Bremner, J., Shobe, K. K., & Kihlstrom, J. F. (2000). False memories in women with self-
        reported childhood sexual abuse: An empirical study. Psychological Science, 11, 333-
        337.
Burns, D. J. (1993). Item gains and losses during hypermnesic recall: Implications for the
        item-specific–relational information distinction. Journal of Experimental Psychology:
        Learning, Memory, and Cognition, 19, 163-173.
Cabeza, R., & Lennartson, E. R. (2005). False memory across languages: Implicit associative
        response vs fuzzy trace view. Memory, 13, 1-5.
Ceci, S. J., & Bruck, M. (1993). Suggestibility of the child witness: A historical review and
        synthesis. Psychological Bulletin, 113, 403-439.
Clancy, S. A., Schacter, D. L., McNally, R. J., & Pitman, R. K. (2000). False recognition in
        women reporting recovered memories of sexual abuse. Psychological Science, 11, 26-
        31.
Coane, J. H., McBride, D. M., Raulerson, B. A. III, & Jordan, J. S. (2006). False memory in a
        short-term memory task. Experimental Psychology, 54, 62-70.
Steffens & Mecklenbräuker                                               False Memories    22

Deese, J. (1959). On the prediction of occurrence of particular verbal intrusions in immediate
        recall. Journal of Experimental Psychology, 58, 17-22.
Dodd, M. D., & MacLeod, C. M. (2004). False recognition without intentional learning.
        Psychonomic Bulletin & Review, 11, 137-142.
Dunning, D., & Sherman, D. A. (1997). Stereotypes and tacit inference. Journal of
        Personality and Social Psychology, 73, 459-471.
Fisher, R. P., & McCauley, M. R. (1995). Improving eyewitness testimony with the cognitive
        interview. In J. R. Graham & M. S. Zaragoza (Eds.), Memory and testimony in the
        child witness (pp. 141-159). Thousand Oaks, CA, US: Sage.
Freyd, J. J., & Gleaves, D. H. (1996). "Remembering" words not presented in lists: Relevance
        to the current recovered/false memory controversy. Journal of Experimental
        Psychology: Learning, Memory, and Cognition, 22, 811-813.
Gallo, D. A. (2006). Associative illusions of memory. False memory research in DRM and
        related tasks. New York, NY, USA: Psychology Press.
Gallo, D. A., Bell, D. M., Beier, J. S., & Schacter, D. L. (2006). Two types of recollection-
        based monitoring in younger and older adults: Recall-to-reject and the distinctiveness
        heuristic. Memory, 2006, 14, 730-741.
Garry, M., & Gerrie, M. P. (2005). When photographs create false memories. Current
        Directions in Psychological Science, 14, 321-325.
Garry, M., Manning, C. G., Loftus, E. F., & Sherman, S. J. (1996). Imagination inflation:
        Imagining a childhood event inflates confidence that it occurred. Psychonomic Bulletin
        and Review, 3, 208-214.
Geraerts, E., Arnold, M. M., Lindsay, D. S., Merckelbach, H., Jelicic, M., & Hauer, B.
        (2006). Forgetting of prior remembering in persons reporting revocered memories of
        childhood sexual abuse. Psychological Science, 17, 1002-1008.
Geraerts, E., Smeets, E., Jelicic, M., van Heerden, J., & Merckelbach, H. (2005). Fantasy
        proneness, but not self-reported trauma is related to DRM performance of women
        reporting recovered memories of childhood sexual abuse. Consciousness and
        Cognition, 14, 602-612.
Gleaves, D. H., Smith, S. M., Butler, L. D., & Spiegel, D. (2004). False and recovered
        memories in the laboratory and clinic: A review of experimental and clinical evidence.
        Clinical Psychology: Science and Practice, 11, 3-28.
Goff, L. M., & Roediger, H. L., III. (1998). Imagination inflation for action events: Repeated
        imaginings lead to illusory recollections. Memory and Cognition, 26, 20-33.
Hirt, E. R., Lynn, S., Payne, D. G., Krackow, E., & McCrea, S. M. (1999). Expectancies and
        memory: Inferring the past from what must have been. In I. Kirsch (Ed.), How
        expectancies shape experience (pp. 93- 124). Washington, DC, USA: American
        Psychological Association.
Hunt, R. R., & Einstein, G. O. (1981). Relational and item-specific information in memory.
        Journal of Verbal Learning and Verbal Behavior, 20, 497-514.
Hutchison, K. A., & Balota, D. A. (2005). Decoupling semantic and associative information
        in false memories: Explorations with semantically ambiguous and unambiguous
        critical lures. Journal of Memory and Language, 52, 1-28.
Hyman, I. E., Husband, T. H., & Billings, F. (1995). False memories of childhood
        experiences. Applied Cognitive Psychology, 9, 181-197.
Jacoby, L. L., Kelley, C. M., & Dywan, J. (1989). Memory attributions. In H. L. Roediger III
        & F. I. M. Craik (Eds.), Varieties of memory and consciousness: Essays in honour of
        Endel Tulving (pp. 391-422). Hillsdale, NJ, USA: Lawrence Erlbaum Associates.
James, W. (1890). Principles of psychology (Vol. I). London: MacMillan.
Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological
        Bulletin, 114, 3-28.
You can also read